Electroconvulsive therapy (ECT) is an extremely effective treatment for major depression and patients that receive this modality frequently present with the most severe, recurrent forms of this illness. Perhaps the most critical clinical issue facing the field of ECT is the problem of early relapse. In unipolar patients who have responded to ECT, the current standard of practice is to use tricyclic antidepressant (TCA) continuation therapy in the prevention of relapse. This practice is based largely on the findings of three controlled trials that were conducted in England in the 1960's. Besides serious methodological flaws, the relevance of these studies to present practice is questionable. In this early work, ECT was frequently used as a 'first choice' treatment and standards for adequate pharmacological treatment have changed considerably over this period. In current practice, resistance to adequate trials of antidepressant medications is a primary indication for ECT and medication-resistant patients form a large proportion of ECT samples. Consequently, it is common to use as continuation therapy following ECT the very same class of antidepressant medications that patients failed during treatment of the acute episode. The efficacy of this practice has never been substantiated. Indeed, in a preliminary prospective, naturalistic study, we found that the relapse rate was twice as high in patients who had failed one or more adequate TCA trials prior to ECT, compared to patients who came to ECT without receiving an adequate medication trial. Adequacy of postECT TCA pharmacotherapy was only marginally related to relapse rates, with the patients who benefitted appearing to be only those who had not failed adequate antidepressant treatment prior to ECT. In related research, we have also found that medication resistance appears to be a strong predictor of ECT outcome. Patients who failed adequate TCA trials prior to ECT had a lower ECT response rate. Here we propose to re-evaluate the utility of continuation pharmacotherapy in ECT responders. In a multi-center trial, involving the Carrier Foundation, New York State Psychiatric Institute, and Western Psychiatric Institute and Clinic, a parallel group, random assignment, double-blind design will be used to establish the relative efficacies of placebo, nortriptyline, and combination nortriptyline-lithium carbonate continuation therapies in the prevention of relapse in unipolar patients following response to ECT. It is hypothesized that nortriptyline alone or its combination with lithium will be effective in patients without documented medication resistance, but that only the combination continuation treatment will be efficacious in patients who have failed one or more trials of a cyclic antidepressant prior to ECT. In addition, a larger sample of patients will be prospectively evaluated with respect to clinical features and treatment history, with standardization of ECT administration across sites. The hypothesis will be tested that medication resistance is also a potent predictor of ECT efficacy, and, when considered, is responsible for the apparent association of better ECT response in depressed patients with psychosis.